How can I find the value in matrix

I have a matrix:

M = 
    NaN   23              37              53             53
    NaN 42308.00000  42308.00019  42308.00037  42308.00056
    14  0.2998          0.4497          0.0996          0.2495
    23  0.2002          0.0503          0.4004          0.2505
    48  0.7998          0.9497          0.5996          0.7495
    53  0.7002          0.5503          0.9004          0.7505

The first row is the name of columns and the first columns is the names of rows, How can I find the value in M where its column label and row label matched for the whole matrix, the final output should be vector of the values.

The output is the value for example column head 23 and row name 23 I want to get the value 0.2002, and so on for all matches.

          23                 37             53              53
       42308.00000               42308.00037  42308.00056
    14  
    23  0.2002  
    48            
    53                                    0.9004            0.7505

So the output as a 2 vector2 of the following: [0.2002 0.9004 0.7505] and [42308.00000 42308.00037 42308.00056] I am using the following useful script:

[tf,loc] = ismember(M(2:end,1), M(1,2:end));
output = M( sub2ind(size(M), find(tf)+1, loc(tf)+1) );

But, there is no repetition with the row label, while the column label can have repetition, for ex. may have 2 columns label with the same label as 53 could 4 columns with the same label. Many thanks for help!

[Guillaume]

As described, M cannot be constructed in matlab. Can you describe your M using valid matlab syntax?

[Rafat Qubaja]

I think I fixed it! Thanks

Answer 1

[Rik]

The code below should do it.

M=[NaN   23      37      53       53
    14  0.2998  0.4497  0.0996  0.2495
    23  0.2002  0.0503  0.4004  0.2505
    48  0.7998  0.9497  0.5996  0.7495
    53  0.7002  0.5503  0.9004  0.7505];
mc=M(1,2:end);
mr=M(2:end,1);
L= mc==mr;%requires implicit expansion
% If you have an older release, use this:
% L= bsxfun(@eq,mc,mr)
%extend L to match the size of M
L=[false(1,size(M,2));
    false(size(L,1),1) L];
wanted_vector=M(L)';

[Rafat Qubaja]

Many thanks for the answer Can you help me to have other vector that I add it with time as I did in the M, since its time.

[Rik]

Change the way mr is defined in my code to exclude all rows that start with NaN, then use sum or any on L to find the columns that have a hit.

Also, instead of editing your question, please post the new situation as a comment. With your edit my answer becomes less clear.

[Rafat Qubaja]

Sorry for that since I am not experit with this website, so I thought easier to interactive by re edit the question. Your script is excellent, but I did not manage to extract the 2 vectors from the matrix to get the following vectors Sir can you help me with that, I appreciate that: [0.2002 0.9004 0.7505] and [42308.00000 42308.00037 42308.00056]

[Rik]

Have you tried my suggestion from the first part of my previous comment?

If you want help on this forum, don't ignore the frequent contributors. Your other question was closed for being a duplicate. If you don't want it to be a duplicate, edit it in a meaningful way, and link back to this question.

[Rafat Qubaja]

Again, I am sorry for that and I don't mean anything bad, All of the contributors very nice with me. Ok I will continue here and stope the others, I tried your script, it is excellent to get the first vector, but for the second output I did not manged to get it, Can I get some help on that? Cheers

[Rafat Qubaja]

I deleted all duplicate, and I am linked here, I hope to get help. Thanks

[Rik]

"Change the way mr is defined in my code to exclude all rows that start with NaN, then use sum or any on L to find the columns that have a hit."

That should at least help you part of the way. Try it out, and if you don't understand what I mean, or get any errors, post what you have tried and we can try to help you.

[Rafat Qubaja]

If I will exclude the second row which also start with NaN then this will not be useful, since it contains the data which I look for it and start with nan:

   M = 
      NaN     23              37              53             53
      NaN   42308.00000  42308.00019  42308.00037  42308.00056
      14    0.2998          0.4497          0.0996          0.2495
      23    0.2002          0.0503          0.4004          0.2505
      48    0.7998          0.9497          0.5996          0.7495
      53    0.7002          0.5503          0.9004          0.7505

The output should be 2 vector as following: [0.2002 0.9004 0.7505] and the time corresponding to these values from row #2 which is

[42308.00000 42308.00037 42308.00056]

[Rik]

I didn't mean you should edit M, you should exclude all rows starting with NaN from mr, then go through the rest of your code. Then you have a logical matrix L that contains which points you want to use, so you'll have to reduce that to a row. You can do this with the any function.

[Rafat Qubaja]

Thanks for your reply, I think I stuck since I am not experit with matlab. sorry I can't understand.

[Rafat Qubaja]

This is a matrix: M = NaN 23 37 53 53 NaN 2 5 6 7 14 0.2998 0.4497 0.0996 0.2495 23 0.2002 0.0503 0.4004 0.2505 48 0.7998 0.9497 0.5996 0.7495 53 0.7002 0.5503 0.9004 0.7505 The first row is the labels of columns and the first columns is the labels of rows, How can I find the values from the second row where its column label and row label matched for the whole matrix, the final output should be vector of the values from the second row only. The output is the value for example column head 23 and row name 23 I want to get the value 2, and so on for all matches. So the output is a vector as the following: [2 6 7] But, there is no repetition with the row label, while the column label can have repetition, for ex. may have 2 columns label with the same label as 53. Many thanks for help!

[Rik]

You forgot to select your code and push the {}Code button, so your matrix is unreadable. Therefore, I used the matrix in your question body instead. The code below shows a method to get the vectors you want. It will return a normal vector in case there is only 1 vector, and it will return a cell array if there are multiple vectors as output.

I also appended a display module that prints the vectors with 5 decimals. It shows the structure of the result in the command window, so you can ignore that part if you want.

M=[NaN 23 37 53 53
    NaN 42308.00000 42308.00019 42308.00037 42308.00056
    14  0.2998  0.4497  0.0996  0.2495
    23  0.2002  0.0503  0.4004  0.2505
    48  0.7998  0.9497  0.5996  0.7495
    53  0.7002  0.5503  0.9004  0.7505];
mc=M(1,2:end);
mr=M(~isnan(M(:,1)),1);
NaN_rows=sum(isnan(M(:,1)));
L= mc==mr;%requires implicit expansion
% If you have an older release, use this:
% L= bsxfun(@eq,mc,mr)
%extend L to match the size of M
L=[false(size(M,1)-size(L,1),size(M,2));
    false(size(L,1),1) L];
if NaN_rows==1
    wanted_vector=M(L)';
else
    wanted_vector=cell(NaN_rows,1);
    wanted_vector{1}=M(L)';
    L2=any(L,1);
    for n=2:NaN_rows
        wanted_vector{n}=M(n,L2);
    end
end
%display the result:
clc
if isa(wanted_vector,'cell')
    for n=1:numel(wanted_vector)
    FormatSpec=['wanted_vector{' num2str(n) '}=[ ' ...
        repmat('%.5f ',1,numel(wanted_vector{n})) '];\n'];
    fprintf(FormatSpec,wanted_vector{n})
    end
else
    FormatSpec=['wanted_vector=[ ' ...
        repmat('%.5f ',1,numel(wanted_vector)) '];\n'];
    fprintf(FormatSpec,wanted_vector) %#ok<PFCEL>
end

[Rafat Qubaja]

Dear Brother Rik, Many Many thanks for the great job, you did great, mervelas and magic work, If you please let me know how I can say thank you in this website more than words. Cheers, rafat